JP2006350641A - Parallel line detection method and parallel line detector - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a parallel line detection method and a parallel line detector for detecting a straight line parallel to the travelling direction of a vehicle precisely and reducing a time for processing spent for detection. <P>SOLUTION: The parallel line detector for detecting the straight line parallel to the travelling direction of the vehicle by using a photographed video outside the vehicle has: an outbreak point detection means for detecting an optical flow outbreak point in the video; an edge point detecting means for detecting an edge point in the video; an angle calculation means for calculating an angle formed of a first straight line to be a straight line connecting the edge point and the optical flow outbreak point and a second straight line to be a predetermined straight line; and a parallel line detection means for detecting parallel lines from first straight lines on the basis of the distribution of angles. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a parallel line detection method and a parallel line detection apparatus for recognizing a straight line parallel to a traveling direction of a vehicle, such as a white line or a road shoulder, using an image captured by a camera.

  Many vehicles in recent years are equipped with a camera capable of shooting moving images. And the process which acquires the information outside a vehicle from the image | video image | photographed with this camera, provides a driver | operator with required information, or is useful for vehicle control is proposed.

  One example of such processing is white line recognition processing. The white line here is, for example, a white line indicating a center line or a road shoulder, and this recognition processing is very important, for example, for automatic driving of a vehicle.

There exists patent document 1 as what recognizes this white line. Patent Document 1 discloses an embodiment in which white line data D4 is acquired at every time Δt to estimate a white line or use GPS. In addition, another method that recognizes a white line uses a Hough transform.
JP 2000-259998 A

  However, the white line estimation means disclosed in Patent Document 1 cannot be said to have sufficient accuracy of the estimated white line. In addition, white line recognition using the Hough transform has a problem in that it requires a search in a two-dimensional space of angles and positions, and processing time is required.

  The present invention has been made in view of such problems, and an object of the present invention is to provide a parallel line detection method and a parallel line detection device that accurately detect a straight line parallel to the traveling direction of the vehicle and reduce the processing time required for detection. .

  In order to solve the above-mentioned problems, the present invention is a parallel line detection device that detects a straight line parallel to the traveling direction of a vehicle using a photographed image outside the vehicle, and detects an optical flow occurrence point in the image. Generating point detecting means, edge point detecting means for detecting an edge point in the video, a first straight line that is a straight line connecting the edge point and the optical flow generating point, and a first straight line that is predetermined. An angle calculating means for calculating an angle formed by the two straight lines; and a parallel line detecting means for detecting a parallel line from the first straight line based on the distribution of the angles.

  In order to solve the above-mentioned problem, the present invention is directed to the parallel line detecting means that detects a maximal angle that is maximal in the number of distributions of the angle, and the distribution number of the maximal angle is a predetermined number. A maximum angle exceeding a threshold value is acquired, and the first straight line forming the acquired angle is detected as a parallel line.

  In order to solve the above-mentioned problem, the present invention is directed to the parallel line detecting means that detects a maximal angle that is maximal in the number of distributions of the angle, and among the maximal angles, the distribution number is in descending order. A predetermined number of maximum angles are acquired, and the first straight line forming the acquired angle is detected as a parallel line.

  In order to solve the above problems, the present invention provides a parallel line detection method for detecting a straight line parallel to the traveling direction of a vehicle using a photographed image outside the vehicle, and an optical flow generation point in the image A generation point detection step of detecting the edge point, an edge point detection step of detecting an edge point in the video, a first straight line that is a straight line connecting the edge point and the optical flow generation point, and a predetermined straight line An angle calculation step of calculating an angle formed with a second straight line, and a parallel line detection step of detecting a parallel line from the first straight line based on the distribution of the angle.

  In order to solve the above-mentioned problem, in the present invention, the parallel line detection step detects a maximal angle that is maximal in the number of distributions of the angle, and the number of distributions among the maximal angles is a predetermined number. A maximum angle exceeding a threshold value is acquired, and the first straight line forming the acquired angle is detected as a parallel line.

  In order to solve the above-mentioned problem, in the present invention, the parallel line detection step detects a maximal angle that is maximal in the number of distributions of the angle, and the distribution number among the maximal angles is in descending order. A predetermined number of maximum angles are acquired, and the first straight line forming the acquired angle is detected as a parallel line.

  ADVANTAGE OF THE INVENTION According to this invention, the parallel line detection method and parallel line detection apparatus which detected the straight line parallel to the advancing direction of a vehicle correctly, and reduced the processing time concerning a detection can be provided.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In addition, the straight line in this Embodiment shall include a line segment.

  First, an outline of the parallel line detection method in the present embodiment will be described. FIG. 1 is a view showing an image taken outside the vehicle. In this video, an optical flow generation point 10 (hereinafter sometimes referred to as a generation point) and straight lines 11, 12, 13, and 14 are shown. The straight line 11 is a straight line resulting from the projection of the guardrail. The straight lines 12 and 14 are straight lines resulting from the projection of white lines indicating road shoulders. The straight line 13 is a straight line resulting from the projection of the center line. Any of the straight lines 11, 12, 13, and 14 is a straight line parallel to the traveling direction of the vehicle.

  It is known that an extension of a straight line parallel to the traveling direction of such a vehicle includes a generation point 10 as shown in FIG. Therefore, this video is expressed on xy coordinates as shown in FIG. The xy coordinates in the case of FIG. 3 focus only on the straight line 12 and the generation point 10 for easy understanding of the explanation.

  In the xy coordinates, a straight line that is a target for calculating an angle with the x-axis is a straight line that connects the edge point and the generation point 10 in the video. This straight line corresponds to the first straight line. The edge point is mainly a point formed by the edge of a white line or standing tree in a video in which a white line or standing tree is projected.

  In FIG. 3, the coordinates of the generation point 10 are (a, b), and the coordinates of an arbitrary point 15 on the straight line 12 are (c, d). Points 20 and 21 are edge points in an image on which a standing tree is projected. The straight line 12 is a set of projected white line edge points.

  Furthermore, an angle formed by the straight line 12 and the x axis (y = 0) is θ. Further, the angles formed by the straight line connecting the point 20 or 21 and the generation point 10 and the x-axis are φ and ρ, respectively. This y = 0 corresponds to the second straight line.

  In this case, since the angle formed by the straight line connecting all the points on the straight line 12 and the generation point 10 (that is, the straight line 12) and the x-axis is θ, as shown in FIG. It can be said that it is more than the number of distributions of φ and ρ. Therefore, it can be said that a straight line forming an angle with a large number of distributions is a parallel line. FIG. 4 shows the number of angular distributions in FIG. In FIG. 2, since four straight lines are drawn, four maximum values are shown.

  The above is the outline of the parallel line detection method in the present embodiment. Hereinafter, a parallel line detection apparatus that executes the parallel line detection method and processing in the parallel line detection apparatus will be described. In the following description, a straight line connecting an edge point and an optical flow generation point may be simply expressed as a straight line L.

  FIG. 5 is a diagram illustrating the positions of the camera 31 that captures the outside of the vehicle and the parallel line detection device 30 in the vehicle. The camera 31 is attached so as to be able to photograph the front of the vehicle, and transmits the photographed image to the parallel line detection device 30. The parallel line detection device 30 is incorporated in, for example, an ECU and detects parallel lines from an image transmitted by the camera 31. If it is desired to detect parallel lines at the rear of the vehicle, a camera is installed at a position where the rear of the vehicle can be photographed.

  Next, a block diagram of the parallel line detection device 30 will be described with reference to FIG. The parallel line detection device 30 includes a generation point detection unit 51, an edge point detection unit 52, an angle calculation unit 53, and a parallel line detection unit 34. The parallel line detection device 30 includes a CPU for operating the above-described units, which are software, a ROM that stores programs, and a RAM that develops programs and data.

  The generation point detection unit 51 receives an image from the camera and detects an optical flow generation point in the image. The edge point detection unit 52 receives an image from the camera and detects an edge point in the image. The angle calculation unit 53 obtains a straight line L from the optical flow generation point acquired from the generation point detection unit 51 and each edge point acquired from the edge point detection unit 52. The angle calculation unit 53 calculates an angle formed by the straight line L corresponding to each edge point and a predetermined straight line.

  The predetermined straight line may be any straight line, but when calculating the angle, the same straight line is used for all the straight lines L. In FIG. 3 described above, the predetermined straight line is y = 0.

  Here, an example of the angle calculation method of the angle calculation unit 53 will be described taking the case of FIG. 3 as an example. The angle calculation unit 53 calculates (b−d) / (ac). The value obtained by this calculation is the slope of the straight line 12. Therefore, the angle can be calculated by obtaining θ satisfying (b−d) / (ac −) = tan θ. A function for obtaining an arc tangent is provided as a standard function in a high-level language such as C language.

  As shown in the angle calculation method example described above, as a result, the angle formed by the straight line L and y = 0 is calculated.

  Next, the parallel line detector 34 will be described. The parallel line detector 34 detects a parallel line from the straight line L based on the distribution of angles. And the parallel line detection part 34 outputs the parallel line information regarding a parallel line. Since the parallel line information only needs to be able to identify the parallel line, for example, the parallel line information may be composed of two pieces of information of the y-intercept of the parallel line in the xy coordinates and the angle calculated by the angle calculation unit. As described above, the parallel line information can be used for automatic driving of the vehicle.

  There are two types of parallel line detection methods. In the first method, the parallel line detection unit 34 detects a maximum angle at which the number of angular distributions is maximum, and among these maximum angles, the maximum angle at which the number of distributions exceeds a predetermined threshold is determined. This is a method of acquiring and detecting a straight line L forming the acquired angle as a parallel line. According to this method, the number of detected parallel lines is indefinite, but there is a high possibility that all parallel lines can be detected.

  Next, in the second method, the parallel line detection unit 34 detects the maximum angle that is the maximum in the number of distributions of the angle, and the maximum number of maximums in the order of the distribution number among the maximum angles. This is a method of acquiring an angle and detecting a straight line L forming the acquired angle as a parallel line. According to this method, the upper limit of the number of detected parallel lines is determined, but the possibility that all parallel lines can be detected is reduced. However, since the vehicle normally travels between two white lines, if the predetermined number is determined to be two to three, there is a high possibility that only effective white lines can be detected.

  Next, the processing of the parallel line detection apparatus will be described using the flowchart of FIG. This flowchart is a flowchart showing a process from when an image is input to the parallel line detection apparatus to when a parallel line is detected.

  In step S101, the generation point detection unit detects an optical flow generation point from the video. Next, in step S102, the edge point detection unit detects each edge point from the video. In step S103, the angle calculation unit 53 calculates an angle formed by the optical flow generation point, the straight line connecting each edge point, and the x axis. In step S104, the parallel line detection unit detects parallel lines from the straight line L based on the distribution of angles.

  As described above, the parallel line detection method in the present embodiment can reduce the processing time required for detection because the calculation amount is very small compared to the case where the Hough transform is used. A straight line parallel to the traveling direction of the vehicle, such as a white line, can be detected.

It is a figure which shows the image | video outside the image | photographed vehicle. It is a figure which shows that the optical flow generation | occurrence | production point is contained in the extended straight line. It is a figure which shows xy coordinate. It is a figure which shows the distribution number of an angle. It is a figure which shows the position in the vehicle of a camera and a parallel line detection apparatus. It is a block diagram of a parallel line detection apparatus. It is a flowchart which shows the process of a parallel line detection apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Optical flow generation | occurrence | production point 11, 12, 13, 14 Straight line 15, 20, 21 Point 30 Parallel line detection apparatus 31 Camera 51 Generation point detection part 52 Edge point detection part 53 Angle calculation part 54 Parallel line detection part

Claims (6)

A parallel line detection device that detects a straight line parallel to the traveling direction of the vehicle using a photographed image outside the vehicle,
Generation point detecting means for detecting an optical flow generation point in the video;
Edge point detection means for detecting an edge point in the video;
Angle calculation means for calculating an angle formed by a first straight line that is a straight line connecting the edge point and the optical flow generation point, and a second straight line that is a predetermined straight line;
A parallel line detection device comprising: parallel line detection means for detecting a parallel line from the first straight line based on the distribution of the angles. The parallel line detection means detects a maximum angle that is a maximum in the number of distributions of the angle,
Among the maximum angles, obtain a maximum angle whose distribution number exceeds a predetermined threshold,
The parallel line detection apparatus according to claim 1, wherein the first straight line forming the acquired angle is detected as a parallel line. The parallel line detection means detects a maximum angle that is a maximum in the number of distributions of the angle,
Among the maximum angles, a predetermined number of maximum angles are acquired in descending order of the number of distributions,
The parallel line detection apparatus according to claim 1, wherein the first straight line forming the acquired angle is detected as a parallel line. A parallel line detection method for detecting a straight line parallel to the traveling direction of the vehicle using a photographed image outside the vehicle,
A generation point detection stage for detecting an optical flow generation point in the video,
An edge point detecting step for detecting an edge point in the video;
An angle calculation step of calculating an angle formed by a first straight line that is a straight line connecting the edge point and the optical flow generation point and a second straight line that is a predetermined straight line;
A parallel line detecting step of detecting a parallel line from the first straight line based on the distribution of the angles. The parallel line detection step detects a maximum angle that is a maximum in the number of distributions of the angle,
Among the maximum angles, obtain a maximum angle whose distribution number exceeds a predetermined threshold,
The parallel line detection method according to claim 4, wherein the first straight line forming the acquired angle is detected as a parallel line. The parallel line detection step detects a maximum angle that is a maximum in the number of distributions of the angle,
Among the maximum angles, a predetermined number of maximum angles are acquired in descending order of the number of distributions,
The parallel line detection method according to claim 4, wherein the first straight line forming the acquired angle is detected as a parallel line.

Images